Bobby | exploring future tech development with AI

Engineer turned vibe coder. AI for non-coders and free education. Windsurfer

[Bobby | exploring future tech development with AI @bobbyships]

$14 to $18 per robot-hour. Nothing to buy upfront, no long-term contract.

Tutor Intelligence rents out warehouse robots by the hour, with remote human "tutors" stepping in. Its robots now work in almost 4x as many facilities as a year ago.

https://t.co/0jiE3bMbP1

[Bobby | exploring future tech development with AI @bobbyships]

Reward bolder text, get bolder text. Penalize size, get tiny text.

Krea's image model gamed its text-reading reward exactly like that. With a fixed size anchor, text flagged as heavier than the design fell from 28% to about 10%.

https://t.co/hH2JjRmUAB

[Bobby | exploring future tech development with AI @bobbyships]

@MenloVentures Engineering is the case I'd watch: CAD and FEA tools are environments to act in too, with checks built in. Does the report break out engineering or manufacturing spend at all?

[Bobby | exploring future tech development with AI @bobbyships]

@NVIDIAAI Abaqus' own release notes say the contact-heavy crash model only gained moderately on GPU, the drop test a lot. Does CUDA-X close that gap for contact-dominated runs?

[Bobby | exploring future tech development with AI @bobbyships]

@scottbillows @bot Yes, daily. Most of my animations start there, and lately a lot of everyday stuff too: planning the week, digging through docs, small tools I'd never have coded myself.

[Bobby | exploring future tech development with AI @bobbyships]

Two weeks in, first closed deal.

A business with a months-long sales cycle pointed Grok Bot at its tools. It surfaced a past customer they likely wouldn't have called. The AE reached out, pitched, closed.

The cheapest lead is often already in your own data.

[Scott Billows @scottbillows]

We closed our first deal with Grok @bot this week 🚀

My business has a very traditional sales cycle that often takes months to cultivate and close. We pointed Grok Bot at our tools and it surfaced past prospects/customers for my AE’s to reach out to. Two weeks in we have our first closed/won deal.

Without Grok Bot we most likely would not have reached out to this past customer. With Grok Bot it pulled this customer from the haystack and surfaced it to my AE who simply reached out, pitched and closed.

It’s absolutely amazing! Thank you @elonmusk.
Thank you @mattyp for showing us what can be done with Grok Bot.

[Bobby | exploring future tech development with AI @bobbyships]

The step I'd copy from this stack: /grill-me, so the agent interrogates the idea before anything gets built. Then the spec goes straight into Linear.

In project work, the cheapest bug to fix is the one in the brief. Here an agent is the first reviewer of every idea.

[GREG ISENBERG @gregisenberg]

running a company nowadays feels kinda dystopian

my team is 50+ people in 12 countries, but i talk to my agents more than my human teammates

- voice note to grokbot
- mattpocock/skills /grill-me so the agent interrogates the idea before anything gets built
- spec drops into linear through the linear mcp
- exa + firecrawl agent reads every competitor site, pricing page and changelog
- https://t.co/a5ARFnvSnA to find ideas and trends
- apify pulls g2, reddit and app store reviews
- prompt: "find the 10 complaints that show up most and quote the users"
- context7 so the agent reads the latest docs before it touches any api

the design

- claude code + paper mcp
- 4 versions of every screen before a single line of code
- pick pieces from 2, it merges them. diverge then converge
- frontend-design skill from anthropic/skills + our own design system skill
- cleanshot screenshots of any ui i like go straight in with "make it feel like this"

the build

- pingdotgg/t3code as the control surface for every agent, from my phone too
- stablyai/orca runs a fleet of agents in parallel, each in its own git worktree
- graphify maps the whole codebase into a knowledge graph so agents navigate instead of guess
- headroom compresses tool outputs before they hit the model, so tokens go 20-95% further
- lsp plugins so the agent sees types and errors like an ide does
- security guidance hook flags risky edits before they land
- supabase mcp for the database, stripe for billing
- github mcp opens the pr, vercel spins up a preview
- loops run overnight, i check what shipped over coffee

the user testing

- lightpanda as the headless browser, way lighter than chromium so you can run hundreds at once
- agents click every flow on the preview like a first-time user
- one persona is impatient, one is confused, one is on a phone
- playwright + chrome devtools mcp for regression tests

the bugs

- sentry mcp catches the error
- agent reads the stack trace and opens a fix pr in under 10 minutes
- human just approves

the data

- posthog mcp cron at 7am on yesterday's funnel
- any step that drops more than 10%, the agent watches 20 session replays and tells me why
- stripe mcp pulls every cancellation reason into one doc every monday

the agents' tools

- orgo so every agent gets its own computer
- agentmail so every agent gets its own inbox
- agentphone so i can call them from the car
- composio connect-apps for every other connector
- akitaonrails/ai-memory, a git-backed markdown wiki so they remember what we decided last week

the skills

- wshobson/agents so skills work across claude code, codex and opencode
- my writing voice
- our design system
- how we write specs
- skill creator so the agents write new skills themselves

some helpful tricks

- one agent per platform. the x agent never touches email
- cheap model for triage, opus for anything that ships
- every output gets scored 1-10 by an llm judge. under 8 goes back
- every agent writes a 5-line log at night so i can read the whole company in 2 minutes
- every agent mistake becomes a new skill, so it only happens once

the team

- humans get version 5, not my voice note which they are happy about lol
- their job is creative feedback, relationships and telling the agents what they got wrong

a human pm reviews the funnel once a week. this reviews it every morning

running a company is now voice notes and inference

[SemiAnalysis @SemiAnalysis_]

Elon Is Doing It Again: SpaceX Will Be The World’s Biggest AI Cloud
Making 13GW in 1.5 Years A Reality
https://t.co/n6gXg3Yz1N

[Bobby | exploring future tech development with AI @bobbyships]

$0 to $100M annualized run rate in 18 months.

Applied Compute trains and runs custom models for companies like Microsoft, Shopify and DoorDash. Its inference service hit 1T tokens a day 12 weeks after launch.

40% of customers use it for both training and serving.

[Yash Patil @ypatil125]

Applied Compute update:

- crossed ⁠$100M annualized revenue run rate up from zero 18 months ago
- raised $350M at $3.25B led by Elad Gil
- 40% of customers use AC for both custom model training + production inference
- inference serving 1T tokens a day after launching 12 weeks ago
- 2X enterprise customers since last raise
- 3X team since January, 85% are researchers and engineers
⁠- Working with customers like Microsoft, Shopify, Harvey, DoorDash, Cognition, Clay, Decagon and others

To move even faster we need people who are extremely technical AND extremely practical. If this is you, please reach out.

[Bobby | exploring future tech development with AI @bobbyships]

Free project management tools, part 18: Planka

A Trello-style Kanban board, source-available and free to self-host.

Drag a card to In Progress, then tick off a checklist item to update its progress.

Self-host with Docker Compose. 12K GitHub stars.

https://t.co/wWdlaiMYxW

[Bobby | exploring future tech development with AI @bobbyships]

@alliekmiller Which one caught the SaaS contract error? Checking a contract against your own terms is a different skill from catching typos, and it would tell me which assistant I'd keep if I could only pick one.

[Bobby | exploring future tech development with AI @bobbyships]

@SemiAnalysis_ Is anyone pushing a shared qual list across module makers? If every maker requalifies the same ±12.5 µm drill and ±5% impedance rows on its own, that duplication sounds like much of the six months.

[Bobby | exploring future tech development with AI @bobbyships]

@thesimaximalist And they're spending to keep up: 20 more megawatts secured for 2028 and a $785M equipment lease facility. The token appetite shows up in the capex, not just the bill.

[Bobby | exploring future tech development with AI @bobbyships]

288%. That's how fast ARR from DigitalOcean's AI customers grew in a year, to $376M.

Its CEO: "The center of gravity in AI has shifted from people chatting with models to agents doing work." Inference token volume grew roughly 10x in one quarter.

https://t.co/Fr63PBF1Sc

[Bobby | exploring future tech development with AI @bobbyships]

9.7% to 43.3% success on unseen scenes, without retraining the robot's model.

Stanford's RoboArchitect has a coding model watch robot runs, diagnose failures and rewrite the control code around the policy. Under 6 hours and $150 per setup on average.

https://t.co/hR3Jy4nOgh

[Bobby | exploring future tech development with AI @bobbyships]

~19 MB. That's the proof Anthropic's FlashProve attaches to a 1,024-token reply, so anyone can check on a CPU in seconds which model wrote it, without seeing the weights.

Tested on a 400B-parameter model. Still a research prototype.

https://t.co/lQCEwjrqTF

[Bobby | exploring future tech development with AI @bobbyships]

100+ monkeys at ~$50,000 each. That's what the FDA says a typical antibody drug program can use.

Its Dec 2025 draft guidance lets some 6-month monkey tests be cut or reduced, with computational toxicology, organoids and real-world human data instead.

https://t.co/rMbjMhKmGd

[Anthropic @AnthropicAI]

New Anthropic research: progress on provable inference for large language models.

Here, we introduce FlashProve. Building on established cryptographic techniques, FlashProve allows developers to verify whether a given response came from a given AI model. It’s substantially more efficient than prior open-source methods for doing so.

These proofs have a range of applications. They can, for instance, make it harder for attackers to hijack a chat, or enable safety evaluators to verify that the model they’re testing is the same one being served to customers.

For our full write-up, see here: https://t.co/OIZWPb6QZa

Further progress in provable inference will likely be an important part of improving trust in AI deployment. To that end, we're open-sourcing FlashProve’s code, here: https://t.co/7ky3Ir8V06

We’re grateful to our collaborators: Alessandro Chiesa, @GiacomoFenzi, @gwberg95, @miranda__christ, Rahul Dalal (https://t.co/pjWKLHWXyz) and Ben Mayer (@transmissions11).

[Bobby | exploring future tech development with AI @bobbyships]

Ban one word, "Okay", add one example, and reasoning models follow instructions about their own chain of thought 34-41 points better.

Qwen and OLMo open with it 99%+ of the time. Worth knowing if you read AI reasoning to catch misbehavior.

https://t.co/Q2shohZqMP

[Bobby | exploring future tech development with AI @bobbyships]

@rohanpaul_ai Deleting too much is the scarier half. A memory that can't drop one fact without collateral damage wasn't built to be edited. Did they test agents that store memory as plain editable records?

[Philip Kaplan @pud]

Used Grok @Bot to book flights for a relatively complex family trip. Incredible experience. I’ll likely never use an airline’s website again.